Objectives Canopy sensors have been widely used for nitrogen nutrition diagnosis of cereal crops recently, however fewer in vegetable crops. We studied its accuracy in nitrogen nutrition and growth diagnosis during vegetable growth seasons and the path of improvement of the accuracy.

Methods A field experiment was conducted with lettuce (Lactuca sativa L.), a popular winter vegetable in the Taihu Lake region, as tested crop materials. Five chemical N rates of 0, 108, 162, 216, 270 kg/hm² were applied. At the seedling, rosette, stem formation and harvest stages, the values of normalized difference vegetation index (NDVI) and the ratio vegetation index (RVI) were determined using GreenSeeker canopy sensor, and the biomass and N content of lettuces were analyzed at the same time. The applicability and predicting accuracy of biomass and N status by canopy sensor were verified.

Results The NDVI and RVI values were significantly related to the aboveground biomass (AGB), root-shoot ratio (RTS), plant N uptake (PNU) and plant nitrogen concentration (PNC) during growth seasons of lettuce, and the NDVI measurements showed higher accuracy indeed. The NDVI values were positively correlated with AGB and PNU, with the coefficients ranged from 0.779 to 0.945 and 0.819 to 0.938, respectively, and negatively correlated with RTS and PNC with the correlation coefficient ranged from –0.367 to –0.844 and –0.328 to –0.732. At the four growth stages, the measured NDVI values at the rosette stage and the stem formation stage had higher prediction accuracy, and the prediction accuracy for AGB, RTS, PNU and PNC was 0.76 to 0.92, 0.37 to 0.71, 0.77 to 0.88 and 0.34 to 0.54, respectively. Using two-year sensor measurements to establish a unified prediction equation for lettuce biomass and nitrogen nutrition status in each growth stage, the rosette stage equation gave the most accurate prediction of AGB (73%), RTS (48%), PNU (52%) and PNC (31%). Based on the whole growth stage^,s prediction equations, the NDVI had high prediction accuracy for AGB, RTS, PNU and PNC, with accurate rate of 54%, 43%, 57% and 26% in turn. After calibrated the equation parameters with index of days after transplantation (DAT), the prediction accuracy rate for the AGB, PNU and PNC was increased to 62%, 71% and 34%, respectively.

Conclusions The active canopy sensor has a great potential for non-destructive diagnosis of biomass and N status of lettuce. The prediction accuracy of aboveground biomass (AGB) and plant N uptake (PNU) based on the NDVI measurement could be as high as 62% and 71%, respectively. After calibration with the days after transplanting, the prediction equation based on the NDVI measurements at whole growing stages could satisfy the accuracy requirement for the fertilization recommendation on low vegetative coverage vegetables like lettuce.